X-Git-Url: https://git.kernel.dk/?a=blobdiff_plain;f=iolog.c;h=51aecd4325e31e62e83dfff629331b376fda3659;hb=e8cf24e570e0bda05dd8f9c390911ebc280956ea;hp=bf293a19090f90110d2f40da684334c2e40fabad;hpb=98e7161cda6a1f5b0cbd8979c261db8eb3a4ab72;p=fio.git

diff --git a/iolog.c b/iolog.c
index bf293a19..51aecd43 100644
--- a/iolog.c
+++ b/iolog.c
@@ -19,10 +19,19 @@
 #include "smalloc.h"
 #include "blktrace.h"
 #include "pshared.h"
+#include "lib/roundup.h"
+
+#include <netinet/in.h>
+#include <netinet/tcp.h>
+#include <arpa/inet.h>
+#include <sys/stat.h>
+#include <sys/socket.h>
+#include <sys/un.h>
 
 static int iolog_flush(struct io_log *log);
 
 static const char iolog_ver2[] = "fio version 2 iolog";
+static const char iolog_ver3[] = "fio version 3 iolog";
 
 void queue_io_piece(struct thread_data *td, struct io_piece *ipo)
 {
@@ -32,18 +41,24 @@ void queue_io_piece(struct thread_data *td, struct io_piece *ipo)
 
 void log_io_u(const struct thread_data *td, const struct io_u *io_u)
 {
+	struct timespec now;
+
 	if (!td->o.write_iolog_file)
 		return;
 
-	fprintf(td->iolog_f, "%s %s %llu %lu\n", io_u->file->file_name,
+	fio_gettime(&now, NULL);
+	fprintf(td->iolog_f, "%lu %s %s %llu %llu\n", utime_since_now(&td->io_log_start_time),
+						io_u->file->file_name,
 						io_ddir_name(io_u->ddir),
 						io_u->offset, io_u->buflen);
+
 }
 
 void log_file(struct thread_data *td, struct fio_file *f,
 	      enum file_log_act what)
 {
 	const char *act[] = { "add", "open", "close" };
+	struct timespec now;
 
 	assert(what < 3);
 
@@ -57,7 +72,9 @@ void log_file(struct thread_data *td, struct fio_file *f,
 	if (!td->iolog_f)
 		return;
 
-	fprintf(td->iolog_f, "%s %s\n", f->file_name, act[what]);
+	fio_gettime(&now, NULL);
+	fprintf(td->iolog_f, "%lu %s %s\n", utime_since_now(&td->io_log_start_time),
+						f->file_name, act[what]);
 }
 
 static void iolog_delay(struct thread_data *td, unsigned long delay)
@@ -108,6 +125,10 @@ static int ipo_special(struct thread_data *td, struct io_piece *ipo)
 
 	f = td->files[ipo->fileno];
 
+	if (ipo->delay)
+		iolog_delay(td, ipo->delay);
+	if (fio_fill_issue_time(td))
+		fio_gettime(&td->last_issue, NULL);
 	switch (ipo->file_action) {
 	case FIO_LOG_OPEN_FILE:
 		if (td->o.replay_redirect && fio_file_open(f)) {
@@ -126,6 +147,11 @@ static int ipo_special(struct thread_data *td, struct io_piece *ipo)
 	case FIO_LOG_UNLINK_FILE:
 		td_io_unlink_file(td, f);
 		break;
+	case FIO_LOG_ADD_FILE:
+		/*
+		 * Nothing to do
+		 */
+		break;
 	default:
 		log_err("fio: bad file action %d\n", ipo->file_action);
 		break;
@@ -134,7 +160,25 @@ static int ipo_special(struct thread_data *td, struct io_piece *ipo)
 	return 1;
 }
 
-static bool read_iolog2(struct thread_data *td);
+static bool read_iolog(struct thread_data *td);
+
+unsigned long long delay_since_ttime(const struct thread_data *td,
+	       unsigned long long time)
+{
+	double tmp;
+	double scale;
+	const unsigned long long *last_ttime = &td->io_log_last_ttime;
+
+	if (!*last_ttime || td->o.no_stall || time < *last_ttime)
+		return 0;
+	else if (td->o.replay_time_scale == 100)
+		return time - *last_ttime;
+
+
+	scale = (double) 100.0 / (double) td->o.replay_time_scale;
+	tmp = time - *last_ttime;
+	return tmp * scale;
+}
 
 int read_iolog_get(struct thread_data *td, struct io_u *io_u)
 {
@@ -143,10 +187,16 @@ int read_iolog_get(struct thread_data *td, struct io_u *io_u)
 
 	while (!flist_empty(&td->io_log_list)) {
 		int ret;
+
 		if (td->o.read_iolog_chunked) {
 			if (td->io_log_checkmark == td->io_log_current) {
-				if (!read_iolog2(td))
-					return 1;
+				if (td->io_log_blktrace) {
+					if (!read_blktrace(td))
+						return 1;
+				} else {
+					if (!read_iolog(td))
+						return 1;
+				}
 			}
 			td->io_log_current--;
 		}
@@ -166,10 +216,11 @@ int read_iolog_get(struct thread_data *td, struct io_u *io_u)
 		io_u->ddir = ipo->ddir;
 		if (ipo->ddir != DDIR_WAIT) {
 			io_u->offset = ipo->offset;
+			io_u->verify_offset = ipo->offset;
 			io_u->buflen = ipo->len;
 			io_u->file = td->files[ipo->fileno];
 			get_file(io_u->file);
-			dprint(FD_IO, "iolog: get %llu/%lu/%s\n", io_u->offset,
+			dprint(FD_IO, "iolog: get %llu/%llu/%s\n", io_u->offset,
 						io_u->buflen, io_u->file->file_name);
 			if (ipo->delay)
 				iolog_delay(td, ipo->delay);
@@ -335,6 +386,9 @@ void trim_io_piece(const struct io_u *io_u)
 
 void write_iolog_close(struct thread_data *td)
 {
+	if (!td->iolog_f)
+		return;
+
 	fflush(td->iolog_f);
 	fclose(td->iolog_f);
 	free(td->iolog_buf);
@@ -342,40 +396,62 @@ void write_iolog_close(struct thread_data *td)
 	td->iolog_buf = NULL;
 }
 
+int64_t iolog_items_to_fetch(struct thread_data *td)
+{
+	struct timespec now;
+	uint64_t elapsed;
+	uint64_t for_1s;
+	int64_t items_to_fetch;
+
+	if (!td->io_log_highmark)
+		return 10;
+
+
+	fio_gettime(&now, NULL);
+	elapsed = ntime_since(&td->io_log_highmark_time, &now);
+	if (elapsed) {
+		for_1s = (td->io_log_highmark - td->io_log_current) * 1000000000 / elapsed;
+		items_to_fetch = for_1s - td->io_log_current;
+		if (items_to_fetch < 0)
+			items_to_fetch = 0;
+	} else
+		items_to_fetch = 0;
+
+	td->io_log_highmark = td->io_log_current + items_to_fetch;
+	td->io_log_checkmark = (td->io_log_highmark + 1) / 2;
+	fio_gettime(&td->io_log_highmark_time, NULL);
+
+	return items_to_fetch;
+}
+
+#define io_act(_td, _r) (((_td)->io_log_version == 3 && (r) == 5) || \
+					((_td)->io_log_version == 2 && (r) == 4))
+#define file_act(_td, _r) (((_td)->io_log_version == 3 && (r) == 3) || \
+					((_td)->io_log_version == 2 && (r) == 2))
+
 /*
- * Read version 2 iolog data. It is enhanced to include per-file logging,
+ * Read version 2 and 3 iolog data. It is enhanced to include per-file logging,
  * syncs, etc.
  */
-static bool read_iolog2(struct thread_data *td)
+static bool read_iolog(struct thread_data *td)
 {
 	unsigned long long offset;
 	unsigned int bytes;
+	unsigned long long delay = 0;
 	int reads, writes, waits, fileno = 0, file_action = 0; /* stupid gcc */
 	char *rfname, *fname, *act;
 	char *str, *p;
 	enum fio_ddir rw;
+	bool realloc = false;
 	int64_t items_to_fetch = 0;
+	int syncs;
 
 	if (td->o.read_iolog_chunked) {
-		if (td->io_log_highmark == 0) {
-			items_to_fetch = 10;
-		} else {
-			struct timespec now;
-			uint64_t elapsed;
-			uint64_t for_1s;
-			fio_gettime(&now, NULL);
-			elapsed = ntime_since(&td->io_log_highmark_time, &now);
-			for_1s = (td->io_log_highmark - td->io_log_current) * 1000000000 / elapsed;
-			items_to_fetch = for_1s - td->io_log_current;
-			if (items_to_fetch < 0)
-				items_to_fetch = 0;
-			td->io_log_highmark = td->io_log_current + items_to_fetch;
-			td->io_log_checkmark = (td->io_log_highmark + 1) / 2;
-			fio_gettime(&td->io_log_highmark_time, NULL);
-			if (items_to_fetch == 0)
-				return true;
-		}
+		items_to_fetch = iolog_items_to_fetch(td);
+		if (!items_to_fetch)
+			return true;
 	}
+
 	/*
 	 * Read in the read iolog and store it, reuse the infrastructure
 	 * for doing verifications.
@@ -384,18 +460,32 @@ static bool read_iolog2(struct thread_data *td)
 	rfname = fname = malloc(256+16);
 	act = malloc(256+16);
 
-	reads = writes = waits = 0;
+	syncs = reads = writes = waits = 0;
 	while ((p = fgets(str, 4096, td->io_log_rfile)) != NULL) {
 		struct io_piece *ipo;
 		int r;
+		unsigned long long ttime;
 
-		r = sscanf(p, "%256s %256s %llu %u", rfname, act, &offset,
-									&bytes);
+		if (td->io_log_version == 3) {
+			r = sscanf(p, "%llu %256s %256s %llu %u", &ttime, rfname, act,
+							&offset, &bytes);
+			delay = delay_since_ttime(td, ttime);
+			td->io_log_last_ttime = ttime;
+			/*
+			 * "wait" is not allowed with version 3
+			 */
+			if (!strcmp(act, "wait")) {
+				log_err("iolog: ignoring wait command with"
+					" version 3 for file %s\n", fname);
+				continue;
+			}
+		} else /* version 2 */
+			r = sscanf(p, "%256s %256s %llu %u", rfname, act, &offset, &bytes);
 
 		if (td->o.replay_redirect)
 			fname = td->o.replay_redirect;
 
-		if (r == 4) {
+		if (io_act(td, r)) {
 			/*
 			 * Check action first
 			 */
@@ -417,7 +507,7 @@ static bool read_iolog2(struct thread_data *td)
 				continue;
 			}
 			fileno = get_fileno(td, fname);
-		} else if (r == 2) {
+		} else if (file_act(td, r)) {
 			rw = DDIR_INVAL;
 			if (!strcmp(act, "add")) {
 				if (td->o.replay_redirect &&
@@ -425,10 +515,9 @@ static bool read_iolog2(struct thread_data *td)
 					dprint(FD_FILE, "iolog: ignoring"
 						" re-add of file %s\n", fname);
 				} else {
-					fileno = add_file(td, fname, 0, 1);
+					fileno = add_file(td, fname, td->subjob_number, 1);
 					file_action = FIO_LOG_ADD_FILE;
 				}
-				continue;
 			} else if (!strcmp(act, "open")) {
 				fileno = get_fileno(td, fname);
 				file_action = FIO_LOG_OPEN_FILE;
@@ -441,7 +530,7 @@ static bool read_iolog2(struct thread_data *td)
 				continue;
 			}
 		} else {
-			log_err("bad iolog2: %s\n", p);
+			log_err("bad iolog%d: %s\n", td->io_log_version, p);
 			continue;
 		}
 
@@ -459,7 +548,9 @@ static bool read_iolog2(struct thread_data *td)
 				continue;
 			waits++;
 		} else if (rw == DDIR_INVAL) {
-		} else if (!ddir_sync(rw)) {
+		} else if (ddir_sync(rw)) {
+			syncs++;
+		} else {
 			log_err("bad ddir: %d\n", rw);
 			continue;
 		}
@@ -470,6 +561,8 @@ static bool read_iolog2(struct thread_data *td)
 		ipo = calloc(1, sizeof(*ipo));
 		init_ipo(ipo);
 		ipo->ddir = rw;
+		if (td->io_log_version == 3)
+			ipo->delay = delay;
 		if (rw == DDIR_WAIT) {
 			ipo->delay = offset;
 		} else {
@@ -480,8 +573,10 @@ static bool read_iolog2(struct thread_data *td)
 			ipo_bytes_align(td->o.replay_align, ipo);
 
 			ipo->len = bytes;
-			if (rw != DDIR_INVAL && bytes > td->o.max_bs[rw])
+			if (rw != DDIR_INVAL && bytes > td->o.max_bs[rw]) {
+				realloc = true;
 				td->o.max_bs[rw] = bytes;
+			}
 			ipo->fileno = fileno;
 			ipo->file_action = file_action;
 			td->o.size += bytes;
@@ -512,12 +607,20 @@ static bool read_iolog2(struct thread_data *td)
 			" read-only\n", td->o.name, writes);
 		writes = 0;
 	}
+	if (syncs)
+		td->flags |= TD_F_SYNCS;
 
 	if (td->o.read_iolog_chunked) {
 		if (td->io_log_current == 0) {
 			return false;
 		}
 		td->o.td_ddir = TD_DDIR_RW;
+		if (realloc && td->orig_buffer)
+		{
+			io_u_quiesce(td);
+			free_io_mem(td);
+			init_io_u_buffers(td);
+		}
 		return true;
 	}
 
@@ -533,16 +636,63 @@ static bool read_iolog2(struct thread_data *td)
 	return true;
 }
 
+static bool is_socket(const char *path)
+{
+	struct stat buf;
+	int r;
+
+	r = stat(path, &buf);
+	if (r == -1)
+		return false;
+
+	return S_ISSOCK(buf.st_mode);
+}
+
+static int open_socket(const char *path)
+{
+	struct sockaddr_un addr;
+	int ret, fd;
+
+	fd = socket(AF_UNIX, SOCK_STREAM, 0);
+	if (fd < 0)
+		return fd;
+
+	addr.sun_family = AF_UNIX;
+	if (snprintf(addr.sun_path, sizeof(addr.sun_path), "%s", path) >=
+	    sizeof(addr.sun_path)) {
+		log_err("%s: path name %s is too long for a Unix socket\n",
+			__func__, path);
+	}
+
+	ret = connect(fd, (const struct sockaddr *)&addr, strlen(path) + sizeof(addr.sun_family));
+	if (!ret)
+		return fd;
+
+	close(fd);
+	return -1;
+}
+
 /*
  * open iolog, check version, and call appropriate parser
  */
-static bool init_iolog_read(struct thread_data *td)
+static bool init_iolog_read(struct thread_data *td, char *fname)
 {
 	char buffer[256], *p;
-	FILE *f;
-	bool ret;
+	FILE *f = NULL;
+
+	dprint(FD_IO, "iolog: name=%s\n", fname);
+
+	if (is_socket(fname)) {
+		int fd;
+
+		fd = open_socket(fname);
+		if (fd >= 0)
+			f = fdopen(fd, "r");
+	} else if (!strcmp(fname, "-")) {
+		f = stdin;
+	} else
+		f = fopen(fname, "r");
 
-	f = fopen(td->o.read_iolog_file, "r");
 	if (!f) {
 		perror("fopen read iolog");
 		return false;
@@ -555,21 +705,24 @@ static bool init_iolog_read(struct thread_data *td)
 		fclose(f);
 		return false;
 	}
-	td->io_log_rfile = f;
+
 	/*
-	 * version 2 of the iolog stores a specific string as the
+	 * versions 2 and 3 of the iolog store a specific string as the
 	 * first line, check for that
 	 */
-	if (!strncmp(iolog_ver2, buffer, strlen(iolog_ver2))) {
-		free_release_files(td);
-		ret = read_iolog2(td);
-	}
+	if (!strncmp(iolog_ver2, buffer, strlen(iolog_ver2)))
+		td->io_log_version = 2;
+	else if (!strncmp(iolog_ver3, buffer, strlen(iolog_ver3)))
+		td->io_log_version = 3;
 	else {
 		log_err("fio: iolog version 1 is no longer supported\n");
-		ret = false;
+		fclose(f);
+		return false;
 	}
 
-	return ret;
+	free_release_files(td);
+	td->io_log_rfile = f;
+	return read_iolog(td);
 }
 
 /*
@@ -593,11 +746,12 @@ static bool init_iolog_write(struct thread_data *td)
 	td->iolog_f = f;
 	td->iolog_buf = malloc(8192);
 	setvbuf(f, td->iolog_buf, _IOFBF, 8192);
+	fio_gettime(&td->io_log_start_time, NULL);
 
 	/*
 	 * write our version line
 	 */
-	if (fprintf(f, "%s\n", iolog_ver2) < 0) {
+	if (fprintf(f, "%s\n", iolog_ver3) < 0) {
 		perror("iolog init\n");
 		return false;
 	}
@@ -617,15 +771,20 @@ bool init_iolog(struct thread_data *td)
 
 	if (td->o.read_iolog_file) {
 		int need_swap;
+		char * fname = get_name_by_idx(td->o.read_iolog_file, td->subjob_number);
 
 		/*
 		 * Check if it's a blktrace file and load that if possible.
 		 * Otherwise assume it's a normal log file and load that.
 		 */
-		if (is_blktrace(td->o.read_iolog_file, &need_swap))
-			ret = load_blktrace(td, td->o.read_iolog_file, need_swap);
-		else
-			ret = init_iolog_read(td);
+		if (is_blktrace(fname, &need_swap)) {
+			td->io_log_blktrace = 1;
+			ret = init_blktrace_read(td, fname, need_swap);
+		} else {
+			td->io_log_blktrace = 0;
+			ret = init_iolog_read(td, fname);
+		}
+		free(fname);
 	} else if (td->o.write_iolog_file)
 		ret = init_iolog_write(td);
 	else
@@ -649,6 +808,7 @@ void setup_log(struct io_log **log, struct log_params *p,
 	INIT_FLIST_HEAD(&l->io_logs);
 	l->log_type = p->log_type;
 	l->log_offset = p->log_offset;
+	l->log_prio = p->log_prio;
 	l->log_gz = p->log_gz;
 	l->log_gz_store = p->log_gz_store;
 	l->avg_msec = p->avg_msec;
@@ -668,16 +828,21 @@ void setup_log(struct io_log **log, struct log_params *p,
 	}
 
 	if (l->td && l->td->o.io_submit_mode != IO_MODE_OFFLOAD) {
+		unsigned int def_samples = DEF_LOG_ENTRIES;
 		struct io_logs *__p;
 
 		__p = calloc(1, sizeof(*l->pending));
-		__p->max_samples = DEF_LOG_ENTRIES;
+		if (l->td->o.iodepth > DEF_LOG_ENTRIES)
+			def_samples = roundup_pow2(l->td->o.iodepth);
+		__p->max_samples = def_samples;
 		__p->log = calloc(__p->max_samples, log_entry_sz(l));
 		l->pending = __p;
 	}
 
 	if (l->log_offset)
 		l->log_ddir_mask = LOG_OFFSET_SAMPLE_BIT;
+	if (l->log_prio)
+		l->log_ddir_mask |= LOG_PRIO_SAMPLE_BIT;
 
 	INIT_FLIST_HEAD(&l->chunk_list);
 
@@ -786,8 +951,8 @@ static void flush_hist_samples(FILE *f, int hist_coarseness, void *samples,
 		entry_before = flist_first_entry(&entry->list, struct io_u_plat_entry, list);
 		io_u_plat_before = entry_before->io_u_plat;
 
-		fprintf(f, "%lu, %u, %u, ", (unsigned long) s->time,
-						io_sample_ddir(s), s->bs);
+		fprintf(f, "%lu, %u, %llu, ", (unsigned long) s->time,
+						io_sample_ddir(s), (unsigned long long) s->bs);
 		for (j = 0; j < FIO_IO_U_PLAT_NR - stride; j += stride) {
 			fprintf(f, "%llu, ", (unsigned long long)
 			        hist_sum(j, stride, io_u_plat, io_u_plat_before));
@@ -804,33 +969,55 @@ static void flush_hist_samples(FILE *f, int hist_coarseness, void *samples,
 void flush_samples(FILE *f, void *samples, uint64_t sample_size)
 {
 	struct io_sample *s;
-	int log_offset;
+	int log_offset, log_prio;
 	uint64_t i, nr_samples;
+	unsigned int prio_val;
+	const char *fmt;
 
 	if (!sample_size)
 		return;
 
 	s = __get_sample(samples, 0, 0);
 	log_offset = (s->__ddir & LOG_OFFSET_SAMPLE_BIT) != 0;
+	log_prio = (s->__ddir & LOG_PRIO_SAMPLE_BIT) != 0;
+
+	if (log_offset) {
+		if (log_prio)
+			fmt = "%lu, %" PRId64 ", %u, %llu, %llu, 0x%04x\n";
+		else
+			fmt = "%lu, %" PRId64 ", %u, %llu, %llu, %u\n";
+	} else {
+		if (log_prio)
+			fmt = "%lu, %" PRId64 ", %u, %llu, 0x%04x\n";
+		else
+			fmt = "%lu, %" PRId64 ", %u, %llu, %u\n";
+	}
 
 	nr_samples = sample_size / __log_entry_sz(log_offset);
 
 	for (i = 0; i < nr_samples; i++) {
 		s = __get_sample(samples, log_offset, i);
 
+		if (log_prio)
+			prio_val = s->priority;
+		else
+			prio_val = ioprio_value_is_class_rt(s->priority);
+
 		if (!log_offset) {
-			fprintf(f, "%lu, %" PRId64 ", %u, %u\n",
-					(unsigned long) s->time,
-					s->data.val,
-					io_sample_ddir(s), s->bs);
+			fprintf(f, fmt,
+				(unsigned long) s->time,
+				s->data.val,
+				io_sample_ddir(s), (unsigned long long) s->bs,
+				prio_val);
 		} else {
 			struct io_sample_offset *so = (void *) s;
 
-			fprintf(f, "%lu, %" PRId64 ", %u, %u, %llu\n",
-					(unsigned long) s->time,
-					s->data.val,
-					io_sample_ddir(s), s->bs,
-					(unsigned long long) so->offset);
+			fprintf(f, fmt,
+				(unsigned long) s->time,
+				s->data.val,
+				io_sample_ddir(s), (unsigned long long) s->bs,
+				(unsigned long long) so->offset,
+				prio_val);
 		}
 	}
 }